harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tim Ellison <t.p.elli...@gmail.com>
Subject Re: [OT]Would anyone help me to verify the patch of harmony-1500 on DRLVM
Date Fri, 29 Dec 2006 19:58:44 GMT
Tony Wu wrote:
> Hi Tim,
> For user defined classes, I believe that no order should be presumed
> when loading them from classpath. But I have no idea on which way
> these vm's boot classes are loaded. This is done in VMI, I have no
> source code of the IBMVM so that I could not verify how these classes
> are loaded.
> I found some comments in the bootclasspath.properties file. [1]
> [1]
> # Upon initialization the VM orders each property of the form
> 'bootclasspath.<n>'
> # on the numerical value of 'n', and sets the bootclasspath to the
> # correspondingly ordered property values.  The values are taken to be
> # file paths relative to the jre/lib/boot directory.

Both the application classpath and the boot classpath are indeed paths,
to be searched in a linear order to find a required type.

If you get different behavior when moving icu4j to the head of the path
then it implies to me that there is a duplication of some type
definition in the JARs that is being exposed by the ordering.

You can use verbose options to see where the VM find types that are loaded.


View raw message